By February 2021, the overall impact of the COVID-19 pandemic in India had been relatively mild in terms of total reported cases and deaths. Surprisingly, the second wave in early April becomes devastating and attracts worldwide attention. Multiple factors (e.g., Delta variants with increased transmissibility) could have driven the rapid growth of the epidemic in India and led to a large number of deaths within a short period. We aim to reconstruct the transmission rate, estimate the infection fatality rate and forecast the epidemic size. We download the reported COVID-19 mortality data in India and formulate a simple mathematical model with a flexible transmission rate. We use iterated filtering to fit our model to deaths data. We forecast the infection attack rate in a month ahead. Our model simulation matched the reported deaths well and is reasonably close to the results of the serological study. We forecast that the infection attack rate (IAR) could have reached 43% by July 24, 2021, under the current trend. Our estimated infection fatality rate is about 0.07%. Under the current trend, the IAR will likely reach a level of 43% by July 24, 2021. Our estimated infection fatality rate appears unusually low, which could be due to a low case to infection ratio reported in previous study. Our approach is readily applicable in other countries and with other types of data (e.g., excess deaths).

/pdf/forecast-of-the-covid-19-trend-in-india-a-simple-modelling-4uhobdbh40.pdf

Forecast of the COVID-19 trend in India: A simple modelling approach.

There is a raising concern of a higher infectious Omicron BA.2 variant and the latest BA.4, BA.5 variant, made it more difficult in the mitigation process against COVID-19 pandemic. Our study aimed to find optimal control strategies by transmission of dynamic model from novel invasion theory.Based on the public data sources from January 31 to May 31, 2022, in four cities (Nanjing, Shanghai, Shenzhen and Suzhou) of China. We segmented the theoretical curves into five phases based on the concept of biological invasion. Then, a spatial autocorrelation analysis was carried out by detecting the clustering of the studied areas. After that, we choose a mathematical model of COVID-19 based on system dynamics methodology to simulate numerous intervention measures scenarios. Finally, we have used publicly available migration data to calculate spillover risk.Epidemics in Shanghai and Shenzhen has gone through the entire invasion phases, whereas Nanjing and Suzhou were all ended in the establishment phase. The results indicated that Rt value and public health and social measures (PHSM)-index of the epidemics were a negative correlation in all cities, except Shenzhen. The intervention has come into effect in different phases of invasion in all studied cities. Until the May 31, most of the spillover risk in Shanghai remained above the spillover risk threshold (18.81-303.84) and the actual number of the spillovers (0.94-74.98) was also increasing along with the time. Shenzhen reported Omicron cases that was only above the spillover risk threshold (17.92) at the phase of outbreak, consistent with an actual partial spillover. In Nanjing and Suzhou, the actual number of reported cases did not exceed the spillover alert value.Biological invasion is positioned to contribute substantively to understanding the drivers and mechanisms of the COVID-19 spread and outbreaks. After evaluating the spillover risk of cities at each invasion phase, we found the dynamic zero-COVID strategy implemented in four cities successfully curb the disease epidemic peak of the Omicron variant, which was highly correlated to the way to perform public health and social measures in the early phases right after the invasion of the virus. 

/pdf/optimal-control-strategies-of-sars-cov-2-omicron-supported-30tpbbqx.pdf

Optimal control strategies of SARS-CoV-2 Omicron supported by invasive and dynamic models

In September 2021, there was an outbreak of coronavirus disease 2019 (COVID-19) in Xiamen, China. Various non-pharmacological interventions (NPIs) and pharmacological interventions (PIs) have been implemented to prevent and control the spread of the disease. This study aimed to evaluate the effectiveness of various interventions and to identify priorities for the implementation of prevention and control measures.The data of patients with COVID-19 were collected from 8 to 30 September 2021. A Susceptible-Exposed-Infectious-Recovered (SEIR) dynamics model was developed to fit the data and simulate the effectiveness of interventions (medical treatment, isolation, social distancing, masking, and vaccination) under different scenarios. The effective reproductive number (Reff ) was used to assess the transmissibility and transmission risk.A total of 236 cases of COVID-19 were reported in Xiamen. The epidemic curve was divided into three phases (Reff = 6.8, 1.5, and 0). Notably, the cumulative number of cases was reduced by 99.67% due to the preventive and control measures implemented by the local government. In the effective containment stage, the number of cases could be reduced to 115 by intensifying the implementation of interventions. The total number of cases (TN) could be reduced by 29.66-95.34% when patients voluntarily visit fever clinics. When only two or three of these measures are implemented, the simulated TN may be greater than the actual number. As four measures were taken simultaneously, the TN may be <100, which is 57.63% less than the actual number. The simultaneous implementation of five interventions could rapidly control the transmission and reduce the number of cases to fewer than 25.With the joint efforts of the government and the public, the outbreak was controlled quickly and effectively. Authorities could promptly cut the transmission chain and control the spread of the disease when patients with fever voluntarily went to the hospital. The ultimate effect of controlling the outbreak through only one intervention was not obvious. The combined community control and mask wearing, along with other interventions, could lead to rapid control of the outbreak and ultimately lower the total number of cases. More importantly, this would mitigate the impact of the outbreak on society and socioeconomics. 

/pdf/model-based-evaluation-of-transmissibility-and-intervention-jxkvvwkj.pdf

Model-Based Evaluation of Transmissibility and Intervention Measures for a COVID-19 Outbreak in Xiamen City, China

Since the COVID-19 outbreak began in early 2020, it has spread rapidly and threatened public health worldwide. Vaccination is an effective way to control the epidemic. In this paper, we model a SAIM equation. Our model involves vaccination and the time delay for people to change their willingness to be vaccinated, which is influenced by media coverage. Second, we theoretically analyze the existence and stability of the equilibria of our model. Then, we study the existence of Hopf bifurcation related to the two equilibria and obtain the normal form near the Hopf bifurcating critical point. Third, numerical simulations based two groups of values for model parameters are carried out to verify our theoretical analysis and assess features such as stable equilibria and periodic solutions. To ensure the appropriateness of model parameters, we conduct a mathematical analysis of official data. Next, we study the effect of the media influence rate and attenuation rate of media coverage on vaccination and epidemic control. The analysis results are consistent with real-world conditions. Finally, we present conclusions and suggestions related to the impact of media coverage on vaccination and epidemic control.

Mathematical modeling and stability analysis of the time-delayed SAIM model for COVID-19 vaccination and media coverage.

https://doi.org/10.1016/j.idm.2022.07.007

Natural history and cycle threshold values analysis of COVID-19 in Xiamen City, China

Background: It is much valuable to evaluate the comparative effectiveness of the coronavirus disease 2019 (COVID-19) prevention and control in the non-pharmacological intervention phase of the pandemic across countries and identify useful experiences that could be generalized worldwide. Methods: In this study, we developed a susceptible – exposure – infectious – asymptomatic – removed (SEIAR) model to fit the daily reported COVID-19 cases in 160 countries. The time-varying reproduction number (Rt) that was estimated through fitting the mathematical model was adopted to quantify the transmissibility. We defined a synthetic index (IAC) based on the value of Rt to reflect the national capability to control COVID-19 was finally calculated by. Results: The goodness-of-fit tests showed that the SEIAR model fitted the data of the 160 countries well. At the beginning of the epidemic, the values of Rt of countries in the European region were generally higher than those in other regions. Among the 160 countries included in the study, all European countries had the ability to control the COVID-19 epidemic. The Western Pacific region did best in the managing of continuous control of the epidemic, with a total of 73.76% of countries can continuously controlled the COVID-19 epidemic, while only 43.63% of the countries in the European region continuously controlled the epidemic, followed by, the Region of Americas with 52.53% of countries, South-east Asia Region with 48% of countries, African region with 46.81% of countries, and Eastern Mediterranean Region with 40.48% of countries. Conclusion: Large variations in controlling the COVID-19 epidemic existed across countries. The world could benefit the experience of some countries that demonstrated the highest containment capabilities.

/pdf/containing-the-transmission-of-covid-19-a-modeling-study-in-4vtjkm4uj9.pdf

Containing the Transmission of COVID-19: A Modeling Study in 160 Countries.

To bridge the ever increasing gap between deep neural networks' complexity and hardware capability, network quantization has attracted more and more research attention. The latest trend of mixed precision quantization takes advantage of hardware's multiple bit-width arithmetic operations to unleash the full potential of network quantization. However, this also results in a difficult integer programming formulation, and forces most existing approaches to use an extremely time-consuming search process even with various relaxations. Instead of solving a problem of the original integer programming, we propose to optimize a proxy metric, the concept of network orthogonality, which is highly correlated with the loss of the integer programming but also easy to optimize with linear programming. This approach reduces the search time and required data amount by orders of magnitude, with little compromise on quantization accuracy. Specifically, on post-training quantization, we achieve 71.27% Top-1 accuracy on MobileNetV2, which only takes 9 seconds for searching and 1.4 GPU hours for finetuning on ImageNet. Our codes are avaliable at this https URL.

Taisong Jin

Papers

Containing the Transmission of COVID-19: A Modeling Study in 160 Countries.

OMPQ: Orthogonal Mixed Precision Quantization.